Twitter As “Crystal Ball”: Predicting Presidential Elections

Lee Rannals for – Your Universe Online

While scientists have used the game “Madden Football” to predict Super Bowl victors, now they say the microblogging site Twitter might help determine who wins the presidential election.

Researchers have found that those who fall in TV shows based on public voting also coincide with what’s going on in the “tweeting” world.

Paying attention to TV shows like American Idol, the researchers were able to anticipate who would be getting the axe week to week by using Twitter signals as a proxy for the general preference of an audience.

Fabio Ciulla from Northeastern University in Boston and his colleagues used the voting system of these shows as a basic test to assess the power of Twitter in relation to how well it can predict things. They relied on the overlap between Twitter users and show audience to collect data on social behavior on a massive scale.

This isn’t the first time Twitter has been used to try and forecast events, such as stock market behavior and election results.

During the study, the scientists found that Twitter activity during the time span limited to the TV show airing and the voting period following it correlated with the contestants’ ranking. This helped predict the outcome of the votes.

Using this approach, the scientists had a simplified version to analyze complex societal phenomena like political elections. Unlike previous voting systems, Twitter offers an indicator that acts as a proxy for what is occurring around the world in real time.

The team showed that the fraction of tweets that included geo-localized information enabled them to internationally map the fan base of each contestant.

They identified a strong influence by the geographical origin of the votes, which suggests a different outcome to the show, assuming that voting had not been just limited to U.S. voters.

Although the method could theoretically be used to predict whether President Barack Obama will retain his position, or if Mitt Romney may take it, the authors did not offer up a prediction as to who will win this year’s election.

The research was published in the journal EPJ Data Science.